(1) Field
The disclosed methods and systems relate generally to compression of packet headers, and more particularly to packet header compression for lossy channels.
(2) Description of Relevant Art
There is a continuing interest in improving the performance of packet-based networks to handle ever-growing demands on the capability of such networks to carry larger capacities of data. Such packet-based networks comprise intricate inter-connections of network devices that operate based on a diverse collection of protocols that define and control the overall operation of the network. These networking protocols require that data carried over the networks be accompanied by control information, which are often quite extensive and can place non-negligible demands on network resources. Thus, one continuing effort to improve network performance is known as “header compression” and involves encoding and/or condensing network protocol control information to decrease the network resources required by such protocols and to allow the networks to handle greater proportions of data.
In general, a packet contains a control portion (the header) that includes various fields that indicate the manner in which the packet should be handled, and a data portion (the payload) that stores the data being transmitted, such as e-mail data, text messages, voice audio, pictures, or video data. It is possible to compress a packet header because there is often significant redundancy between header fields, both within the same packet and between consecutive packets that are part of the same packet stream. For example, with respect to Transmission Control Protocol (TCP) and Internet Protocol (IP), a header compression method described by V. Jacobson in RFC 1144 indicates that for TCP/IP headers, half of a header is likely to remain constant between consecutive packets in a packet stream. Accordingly, a reduction of one-half in the size of a TCP/IP header can be realized by storing a copy of the latest packet header for a packet stream at a receiver and transmitting to the receiver only the variable portion of a subsequent packet header. The method of Jacobson further compresses a TCP/IP header by realizing that the variable portions often change slowly or minutely. Thus, transmitting the amount of change (called “delta”) rather than the value of the variable portion can result in a further reduction in header size. A receiver can apply a received delta to the variable portion of an uncompressed packet header to produce the variable portion of the next header. The constant portion of this next header can be obtained from the constant portion of the uncompressed header.
As exemplified by the header compression method of Jacobson, a “compressed header” is generally a condensed and/or encoded version of a full packet header. Thus, an individual compressed header in a packet does not convey the same extent of control information as a full header and must rely on a context provided by previous header(s) to derive control information. There are, however, situations where such context may be unavailable, For example, it is commonly known that packets may be re-ordered prior to arrival at a receiver. Because a header is compressed based on a previous header, packet re-ordering prior to decompression may cause a context of previous packets to be unavailable and can result in delays and/or other complications at the receiver. In addition, poor transmission conditions may cause certain packets to be lost, which results in decompression errors if a receiver does not recognize the packet loss and continues to apply received deltas to other packets in place of the lost packets. However, even when a receiver recognizes that re-ordering or packet loss has occurred, it may still be unable to accommodate subsequent packets and may require the transmitter to re-send the lost and/or subsequent packets. Thus, there is a continuing interest in improving header compression technology to accommodate problems such as packet re-ordering and packet loss.